Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Quiescence Explosion

Author: Richard Pijl

Date: 02:30:33 11/26/02

Go up one level in this thread


On November 26, 2002 at 04:43:16, David Rasmussen wrote:

>[D]r4rk1/1p1n1pp1/1bq1bn1p/p1pp4/2P2B2/1NNP2P1/PPQ2PBP/R4RK1 w - - 0 19
>
>Chezzz encountered this position in a game. This position, and some immediately
>before and after it, made Chezzz go nuts in quiescence search:
>
>time         nodes depth    score  pv
>00:00:00.14   52kn 02/02/16 ---------------------------------------------------
>00:00:00.14   53kn 03/03/08 +0.055 19. Nb5 a4 20. Nd2
>00:00:00.17   66kn 03/03/12 +0.100 19. Rae1 Rae8 20. Nb5
>00:00:00.25   88kn 03/04/14 ---------------------------------------------------
>00:00:01.37  631kn 04/04/16 +0.205 19. Rae1 a4 20. cxd5 Nxd5
>                                   21. Nxd5 Bxd5
>00:00:01.78  804kn 04/04/16 ---------------------------------------------------
>00:00:02.08  953kn 05/06/16 +0.205 19. Rae1 a4 20. cxd5 Nxd5
>                                   21. Nxd5 Bxd5
>00:00:02.59 1191kn 05/06/16 +0.277 19. Rfe1 a4 20. Nd2 Rfe8
>                                   21. cxd5 Nxd5 22. Bxd5 Bxd5
>                                   23. Rxe8+ Rxe8 24. Qxa4 Qxa4
>                                   25. Nxa4
>00:00:04.14 2020kn 05/06/18 +0.410 19. Qd2 Rfe8 20. Rae1 Red8
>                                   21. Qe3
>00:00:04.20 2040kn 05/06/18 ---------------------------------------------------
>00:00:06.00 3083kn 06/08/18 +0.370 19. Qd2 Rfe8 20. Rae1 Rec8
>                                   21. Qe3 Bc7
>00:00:07.39 3728kn 06/08/19 ---------------------------------------------------
>00:00:11.06 5684kn 07/10/21 +0.483 19. Qd2 Rfe8 20. Rae1 Kf8
>                                   21. cxd5 Nxd5 22. Nxd5 Bxd5
>                                   23. Rxe8+ Kxe8
>00:00:12.47 6313kn 07/10/21 ---------------------------------------------------
>00:00:28.12   14Mn 08/11/22   6/48 19. Rae1
>
>        Time used: 00:00:30.00
>        nps: 497701     nodes: 14936023 Q-nodes: 14513023 (97%)
>        move order: 88%
>        hash probes: 112554     hash hits: 26981 (23%)  hash usage: 2%
>        pawn hash probes: 14816435      pawn hash hits: 14807848 (99%)
>        EGTB probes: 0  EGTB hits: 0 (0%)
>
>97% Q-nodes!!
>In the other positions near this, it was the same. 97-98%. I don't have the most
>brilliant Q-search, but it's not totally naive either. I don't do checks, and I
>use SEE to order moves and to throw out moves that won't bring the score up
>close to alpha. In usual middlegames, I see percentages of 10%-40%.
>
>So how do your engines do in this positions?
>What can be done to lower this number?
>
>/David

Baron has no problem with this position, although the number of qnodes is a bit
above average (60%, normally around 35-40%).

Baron does do checks in qsearch, as well as SEE for move ordering and pruning
In addition, Baron also stores qsearch results in the hashtable (without a
hashmove to prevent bad move ordering) and uses them for pruning as well.

 ply       time     nodes  score  pv
  5( 6)&      0     38233     31  f1e1 a5a4 c4d5 f6d5 b3d2
  5( 6).      0     68691     31  f1e1 a5a4 c4d5 f6d5 b3d2
  6( 8)&      1     90456     27  f1e1 a5a4 b3d2 f8d8 a1c1 d8e8
  6(10).      2    150838     27  f1e1 a5a4 b3d2 f8d8 a1c1 d8e8
  7(11)&      2    225183     28  f1e1 a5a4 b3d2 a4a3 c4d5 f6d5 b2b3
  7(11).      4    425419     28  f1e1 a5a4 b3d2 a4a3 c4d5 f6d5 b2b3
  8(12)&      5    511255     24  f1e1 a5a4 b3d2 a4a3 c4d5 f6d5 c3d5 e6d5
  8(12)&     16   1383298     33  a1e1 a5a4 c4d5 f6d5 b3d2
  8(12)&     30   2500162     38  a2a4 a8e8 c2d2 e8a8 a1e1
  8(12).     30   2509812     38  a2a4 a8e8 c2d2 e8a8 a1e1
  9(14)&     54   4594582     55  a2a4 f8d8 f1e1 d7f8 c4d5 f6d5 g2d5 e6d5 e1e7
  9(14).     60   5207482     55  a2a4 f8d8 f1e1 d7f8 c4d5 f6d5 g2d5 e6d5
 10(15)&    147  12366860     40  a2a4 f8d8 f1e1 d7f8 f4e3 f8g6 e1e2 d8d6 a1e1
 10(15).    192  16267180     40  a2a4 f8d8 f1e1 d7f8 f4e3 f8g6 e1e2 d8d6 a1e1
Computer's move: a2a4

Nodes:         16267180
Qnodes:         9730431
Moves left: 0
Time:        192 seconds
Clock:     33361 seconds

TB probes:         0
TB hits:           0




This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.